System and method for assessing QRS components and likelihood of response to cardiac resynchronization therapy

Abstract

Various embodiments are described herein for a system and a method for assessing a risk of ventricular arrhythmias for a patient. For example, the method may comprise receiving ECG data obtained from the patient; analyzing the ECG data to detect abnormal QRS peaks; determining the risk of ventricular arrhythmias for the patient based on the detected abnormal QRS peaks; and providing an indication of the risk of ventricular arrhythmias for the patient. The system may be configured to perform this method.

The invention claimed is: 
     
       1. A computer implemented method of assessing the likelihood of response to cardiac resynchronization therapy (CRT) for a patient, the method comprising:
 receiving, by a processing unit, time-domain ECG data obtained from the patient; 
 analyzing, by the processing unit, the time-domain ECG data to detect abnormal QRS peaks using a sliding window; 
 determining, by the processing unit, a QRS peak (QRSp) score by quantifying the abnormal QRS peaks in the sliding window of the time-domain ECG data; 
 determining, by the processing unit, a measure of likelihood of CRT response for the patient based on the QRSp score by inputting the QRSp score into a statistical classification model to obtain a probability of CRT response for that patient; and indicating that the patient is a candidate for CRT based on the measure of likelihood of CRT response, 
 
       wherein the statistical classification model is generated using QRSp scores from patients with heart failure who do not respond to CRT and patients with heart failure who respond to CRT; 
       wherein the measure of likelihood of CRT response is used to identify candidates likely to respond to CRT; and 
       determining a treatment regimen for the patient in response to the indication that the patient is a candidate for CRT, wherein the treatment regime includes performing CRT on the patient. 
     
     
       2. The method of  claim 1 , wherein the method comprises acquiring the ECG data from the patient using a plurality of ECG recording leads. 
     
     
       3. The method of  claim 2 , wherein the act of analyzing the ECG data from a given ECG recording lead comprises:
 generating a local QRS (IQRS) signal from X beats of ECG data; 
 generating a global QRS (gQRS) signal from Y beats of ECG data, where X and Y are integers and the X beats of ECG data are contained in the Y beats of ECG data; and 
 comparing the IQRS signal with the gQRS signal to detect the abnormal QRS peaks in the IQRS signal. 
 
     
     
       4. The method of  claim 3 , wherein the IQRS signal is generated by applying time averaging to unfiltered X beats of preprocessed ECG data, and the gQRS signal is generated by filtering the Y beats of preprocessed ECG data using a smoothing filter and then applying time averaging to the filtered Y beats of ECG data, where the X beats of ECG data is a short data window and the Y beats of ECG data is a larger data window that is at least one order of magnitude larger than the short data window. 
     
     
       5. The method of  claim 3 , wherein the comparing comprises:
 identifying positive and negative peaks in the IQRS and gQRS signals; 
 determining abnormal positive peaks in the IQRS signal by counting the number of positive peaks in the IQRS signal within ±M msec of each positive peak in the gQRS signal while excluding the nearest or greatest amplitude IQRS peak within ±M msec of each positive peak in the gQRS signal; 
 determining abnormal negative peaks in the IQRS signal by counting the number of negative peaks in the IQRS signal within ±M msec of each negative peak in the gQRS signal while excluding the nearest or least amplitude IQRS peak within ±M msec of each negative peak in the gQRS signal; and 
 determining the QRSp score as the total of the abnormal positive peaks and the abnormal negative peaks in the IQRS signal. 
 
     
     
       6. The method of  claim 5 , wherein the QRSp score for the patient is a mean, median or maximum of the QRSp scores for ECG data obtained from at least a portion of the ECG recording leads. 
     
     
       7. The method of  claim 1 , wherein the ECG data comprises several sets of ECG data obtained using different ECG leads and the QRSp score is determined for each set of ECG data. 
     
     
       8. The method of  claim 1 , wherein the ECG data comprises precordial ECG data obtained using precordial leads. 
     
     
       9. The method of  claim 1 , further comprising inputting at least one other clinical variable for the patient into the statistical classification model to obtain the probability of CRT response for that patient. 
     
     
       10. The method of  claim 9 , wherein the at least one other clinical variable includes age, gender, etiology of cardiomyopathy, history of atrial fibrillation, left ventricular ejection fraction (LVEF), presence of left bundle branch block (LBBB) and/or QRS duration (QRSd). 
     
     
       11. The method of  claim 1 , wherein the statistical classification model comprises one of a multivariable regression model, a decision tree model, a neural network model and a support vector machine model. 
     
     
       12. The method of  claim 6 , wherein the QRSp score is the maximum of the QRSp scores for ECG data obtained from the at least a portion of the ECG recording leads. 
     
     
       13. The method of  claim 1 , further comprising:
 receiving subsequent ECG data obtained from the patient following the performance of CRT on the patient; 
 analyzing the subsequent ECG data to detect subsequent abnormal QRS peaks; 
 comparing the number of subsequent abnormal QRS peaks to the number of abnormal QRS peaks; and 
 providing an indication of a change in the number of abnormal QRS peaks based on the comparison. 
 
     
     
       14. The method of  claim 13 , further comprising modifying the patient's treatment regimen in response to the change in the number of abnormal QRS peaks. 
     
     
       15. A system for identifying the likelihood of CRT response for a patient to guide CRT candidate selection, wherein the system comprises:
 an input interface for receiving ECG data obtained from the patient; 
 an output interface for providing a measure of likelihood of CRT response for the patient; and 
 a processing unit coupled to the input and the output interfaces, the processing unit being configured to:
 analyze time-domain ECG data in the ECG data obtained from the patient to detect abnormal QRS peaks using a sliding window; 
 determine a QRS peak (QRSp) score by quantifying the abnormal QRS peaks in the sliding window of the time-domain ECG data; 
 determine a measure of likelihood of CRT response for the patient based on the QRSp score by inputting the QRSp score into a statistical classification model to obtain a probability of CRT response for that patient; and 
 provide an indication that the patient is a candidate for CRT based on the measure of likelihood of CRT response, 
 
 wherein the statistical classification model is generated using QRSp scores from patients with heart failure who do not respond to CRT and patients with heart failure who respond to CRT; 
 wherein the measure of likelihood of CRT response is used to identify candidates likely to respond to CRT; and 
 wherein a treatment regimen for the patient is determined in response to the indication that the patient is a candidate for CRT, wherein the treatment regime includes performing CRT on the patient. 
 
     
     
       16. The system of  claim 15 , wherein the ECG data is acquired from the patient using a plurality of ECG recording leads and the processing unit is configured to analyze the ECG data to detect abnormal QRS peaks from a given ECG recording lead by: generating a local QRS (IQRS) signal from X beats of ECG data; generating a global QRS (gQRS) signal from Y beats of ECG data, where X and Y are integers and the X beats of ECG data are contained in the Y beats of ECG data; and comparing the IQRS signal with the gQRS signal to detect the abnormal QRS peaks in the IQRS signal. 
     
     
       17. The system of  16 , wherein the processing unit is configured to compare the IQRS signal with the gQRS signal to detect the abnormal QRS peaks in the IQRS signal by: identifying positive and negative peaks in the IQRS and gQRS signals; determining abnormal positive peaks in the IQRS signal by counting the number of positive peaks in the IQRS signal within ±M msec of each positive peak in the gQRS signal while excluding the nearest or greatest amplitude IQRS peak within ±M msec of each positive peak in the gQRS signal; determining abnormal negative peaks in the IQRS signal by counting the number of negative peaks in the IQRS signal within ±M msec of each negative peak in the gQRS signal while excluding the nearest or least amplitude IQRS peak within ±M msec of each negative peak in the gQRS signal; and determining the QRSp score as the total of the abnormal positive peaks and the abnormal negative peaks in the IQRS signal. 
     
     
       18. A computer implemented method of assessing the likelihood of response to cardiac resynchronization therapy (CRT) for a patient, the method comprising:
 receiving, by a processing unit, time-domain ECG data obtained from the patient; 
 analyzing, by the processing unit, the time-domain ECG data to detect abnormal QRS peaks using a sliding window; 
 determining, by the processing unit, a QRS peak (QRSp) score by quantifying the abnormal QRS peaks in the sliding window of the time-domain ECG data; 
 determining, by the processing unit, a measure of likelihood of CRT response for the patient based on the QRSp score; 
 assigning a CRT peak threshold based on a number of abnormal QRS peaks; 
 comparing the QRSp score to the CRT peak threshold; and 
 indicating that the patient is a candidate for CRT if the QRSp score is greater than the CRT peak threshold, 
 
       wherein the measure of likelihood of CRT response is used to identify candidates likely to respond to CRT; and a treatment regimen for the patient is determined in response to the indication that the patient is a candidate for CR, wherein the treatment regime includes performing CRT on the patient. 
     
     
       19. The method of  claim 18 , wherein the QRSp score for the patient is a mean, median or maximum of the QRSp scores for ECG data obtained from at least a portion of ECG recording leads. 
     
     
       20. The method of  claim 19 , wherein the QRSp score is the maximum of the QRSp scores; and the CRT peak threshold is at least 7.